抄録
To complete cooperative tasks, body movements are important for avoiding collision and unintentional action-interferences. We aimed to develop a robot platform which would allow us to investigate how body movements influence the distribution of roles between robots and humans during a cooperative task. We propose a cooperative task in which robots and humans must complete a single melody on a single set of handbells. The efficient distribution of roles during the task requires nonverbal communication via body movements. We created a robot which can perform this task with humans and evaluated its efficacy. Our robot is able to 1) detect human partner behavior, 2) shift its own behavioral target according to the detection in real-time, and 3) achieve shifted behavioral target in a variety of ways. Thus we show our robot meets necessary criteria as a platform for studying how body movements influence the distribution of roles in cooperative tasks.
